{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/emoji-powered-representation-learning-for","title":"Emoji-Powered Representation Learning for Cross-Lingual Sentiment Classification","arxiv_id":"1806.02557","date":"2018-06-07","proceeding":null,"authors":["Zhenpeng Chen","Sheng Shen","Ziniu Hu","Xuan Lu","Qiaozhu Mei","Xuanzhe Liu"],"abstract":"Sentiment classification typically relies on a large amount of labeled data.\nIn practice, the availability of labels is highly imbalanced among different\nlanguages, e.g., more English texts are labeled than texts in any other\nlanguages, which creates a considerable inequality in the quality of related\ninformation services received by users speaking different languages. To tackle\nthis problem, cross-lingual sentiment classification approaches aim to transfer\nknowledge learned from one language that has abundant labeled examples (i.e.,\nthe source language, usually English) to another language with fewer labels\n(i.e., the target language). The source and the target languages are usually\nbridged through off-the-shelf machine translation tools. Through such a\nchannel, cross-language sentiment patterns can be successfully learned from\nEnglish and transferred into the target languages. This approach, however,\noften fails to capture sentiment knowledge specific to the target language, and\nthus compromises the accuracy of the downstream classification task. In this\npaper, we employ emojis, which are widely available in many languages, as a new\nchannel to learn both the cross-language and the language-specific sentiment\npatterns. We propose a novel representation learning method that uses emoji\nprediction as an instrument to learn respective sentiment-aware representations\nfor each language. The learned representations are then integrated to\nfacilitate cross-lingual sentiment classification. The proposed method\ndemonstrates state-of-the-art performance on benchmark datasets, which is\nsustained even when sentiment labels are scarce.","url_abs":"http://arxiv.org/abs/1806.02557v2","url_pdf":"http://arxiv.org/pdf/1806.02557v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"emoji-powered-representation-learning-for","repo_url":"https://github.com/sInceraSs/ELSA","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"cross-lingual-sentiment-classification","task_name":"Cross-Lingual Sentiment Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"machine-translation","task_name":"Machine Translation"},{"task_slug":"representation-learning","task_name":"Representation Learning"},{"task_slug":"sentiment-analysis","task_name":"Sentiment Analysis"},{"task_slug":"sentiment-classification","task_name":"Sentiment Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}